Multispectral remote sensing land use classification based on RBF neural network with parameters optimized by genetic algorithm

In view of the problems existing in the application of multispectral image classification algorithms, such as low computational speed, low accuracy and difficult convergence, a multispectral remote sensing image classification method based on RBF neural network with parameters optimized by genetic algorithm (GA) is proposed in this paper. This model uses the multispectral sensing remote data and field measurement data of Dadukou District, Chongqing Municipality, to train and test a RBF neural network, and then uses GA to optimize the parameters of the RBF neural network. Then, multispectral sensing data classification is performed on the basis of the trained RBF neural network model. Moreover, the method is compared with the principal component analysis (PCA), linear discriminant analysis (LDA), RBF neural network, orthogonal matching pursuit (OMP) and support vector machine (SVM) algorithms. Experimental results indicate that the overall classification accuracy of this algorithm is improved about 11% compared to other algorithms, which enhances the accuracy of multispectral image classification effectively.